• Release Highlights 2026.1
    • Small Molecule Discovery Suite
      • ROCS X Floes
      • Large Scale Floes
  • Release Notes, Version 2026.1.1
    • AI Fold Floes Release Notes
      • v0.2.6 April 2026
        • General Notice
        • Documentation Updates
        • Floe Updates
    • Cryptic Pocket Detection Floes Release Notes
      • v0.3.3 April 2026
        • General Notice
        • Minor Changes
    • Generative Design Hit-to-Lead Release Notes
      • v2.8.1 April 2026
        • General Notice
        • Floe Updates
    • Large Scale Floes Release Notes
      • 4.4.2 April 2026
        • General Notice
        • Features
        • Minor Changes
    • Molecular Dynamics Affinity Release Notes
      • v6.5.8 April 2026
        • General Notice
        • Floe Updates
    • Molecular Dynamics Core Package
      • v2.6.1 April 2026
        • General Notice
        • Updates
        • Cube Updates
    • ROCS X Floes Release Notes
      • v1.0.0 April 2026
        • General Notice
    • Snowball Release Notes
      • v0.31.4 April 2026
        • General Notice
        • Cube Updates
    • Structural Biology Floes Release Notes
      • v0.2.0 April 2026
        • General Notice
        • Minor Changes
    • Utility Floes Release Notes
      • v2.1.7 April 2026
        • General Notice
        • Floe Updates
  • Recent Release History
  • Small Molecule Discovery Suite
    • Introduction
    • Module-Level Tutorials for the Small Molecule Discovery Suite
      • Target Exploration Module
        • Target Exploration Overview Tutorial
      • Hit Identification Module
        • Hit Identification Overview Tutorial
      • Lead Optimization Module
        • Lead Optimization Overview Tutorial
    • Target Exploration Module
      • Target Exploration Overview Tutorial
      • Cryptic Pocket Detection
        • Introduction
        • Tutorials
        • OpenEye Cryptic Pocket Detection Floes - Documentation
        • Cryptic Pocket Detection Floes Release Notes
        • Legacy Release Notes
        • Bibliography
      • Protein Modeling Floes
        • Introduction
        • Tutorials
        • Protein Modeling Floes - Documentation
        • Protein Modeling Release Notes
        • Protein Modeling Release Notes
      • Structural Biology Floes
        • Introduction
        • Tutorials
        • OpenEye Structural Biology Floes - Documentation
        • Structural Biology Floes Release Notes
        • Legacy Release Notes
        • Bibliography
      • AI Fold Floes
        • Tutorials
        • How-To Guides
        • AI Fold Floes - Documentation
        • AI Fold Floes Release Notes
        • Release Notes
      • Format Conversion
        • orion-etl - Documentation
        • Format Conversion (ETL) Floes
        • Legacy Release Notes
      • Utility Floes
        • Introduction
        • Tutorials
        • How To Guides
        • Utility Floes - Documentation
        • Utility Floes Release Notes
        • Utility Floes Release Notes
        • Historical Release Notes
    • Hit Identification Module
      • Hit Identification Overview Tutorial
      • ChemInfo Hit ID Floes
        • Introduction
        • OpenEye ChemInfo Hit ID Floes - Documentation
        • ChemInfo Hit ID Release Notes
        • ChemInfo Hit ID Release Notes
      • Generative Design Hit-to-Lead Floes
        • Tutorials
        • Floe Reference Documentation
        • Release Notes
        • Legacy Release Notes
      • Large Scale Floes
        • Tutorials
        • How-to Guides
        • Explanations
        • Floe Reference Documentation
        • Release Notes
        • Legacy Release Notes
      • Large Scale Floes Hit-to-Lead
        • Explanations
        • Floe Reference Documentation
        • Release Notes
        • Legacy Release Notes
      • pKa Prediction Floes
        • Introduction
        • Tutorials
        • Frequently Asked Questions
        • Small Molecule pKa Prediction Floes - Documentation
        • pKa Prediction Floes
        • pKa Prediction Floes
      • Protein Modeling Floes
        • Introduction
        • Tutorials
        • Protein Modeling Floes - Documentation
        • Protein Modeling Release Notes
        • Protein Modeling Release Notes
      • ROCS X Floes
        • Introduction
        • Tutorial - Building and Searching a ROCS X 3D Library With Multi-Stage Floes
        • How-To Guides
        • ROCS X Floes - Documentation
        • ROCS X Floes Release Notes
      • Small Molecule Modeling Floes
        • Introduction
        • Tutorials
        • Small Molecule Modeling - Documentation
        • Small Molecule Modeling Release Notes
        • Small Molecule Modeling Release Notes
      • AI Fold Floes
        • Tutorials
        • How-To Guides
        • AI Fold Floes - Documentation
        • AI Fold Floes Release Notes
        • Release Notes
      • Format Conversion
        • orion-etl - Documentation
        • Format Conversion (ETL) Floes
        • Legacy Release Notes
      • Utility Floes
        • Introduction
        • Tutorials
        • How To Guides
        • Utility Floes - Documentation
        • Utility Floes Release Notes
        • Utility Floes Release Notes
        • Historical Release Notes
    • Lead Optimization Module
      • Lead Optimization Overview Tutorial
      • 3D QSAR Modeling Floes
        • Introduction
        • Theory
        • Tutorials
        • OpenEye 3D QSAR Models - Documentation
        • Benchmark Results
        • 3D QSAR Models
        • Legacy Release Notes
        • Bibliography
      • Classic Lead Optimization Floes
        • Introduction
        • Tutorials
        • Classic Lead Opt Floes - Documentation
        • Classic Lead Optimization Release Notes
        • Classic Lead Optimization Release Notes
      • Generative Design Hit-to-Lead Floes
        • Tutorials
        • Floe Reference Documentation
        • Release Notes
        • Legacy Release Notes
      • Generative Design Lead Optimization Floes
        • Tutorials
        • Floe Reference Documentation
        • Release Notes
        • Legacy Release Notes
      • Large Scale Floes Hit-to-Lead
        • Explanations
        • Floe Reference Documentation
        • Release Notes
        • Legacy Release Notes
      • Machine Learning Model Building Floes
        • Introduction
        • Tutorials
        • How-To Guides
        • Floe Reference Documentation
        • Theory
        • FAQs
        • Bibliography
        • Release Notes
        • Legacy Release Notes
      • Molecular Dynamics Affinity Package
        • Introduction
        • How-To Guide and Tutorials
        • MD Affinity Floes - Documentation
        • Molecular Dynamics Affinity Release Notes
        • Legacy Release Notes
      • Permeability Floes
        • Tutorials
        • OpenEye Permeability Floes - Documentation
        • Release Notes
      • pKa Prediction Floes
        • Introduction
        • Tutorials
        • Frequently Asked Questions
        • Small Molecule pKa Prediction Floes - Documentation
        • pKa Prediction Floes
        • pKa Prediction Floes
      • Protein Modeling Floes
        • Introduction
        • Tutorials
        • Protein Modeling Floes - Documentation
        • Protein Modeling Release Notes
        • Protein Modeling Release Notes
      • Quantum Mechanics Psi4 Floes
        • Psi4 QM Floes - Documentation
        • How-To Guides
        • Frequently Asked Questions
        • Release Notes
        • Bibliography
      • Small Molecule Modeling Floes
        • Introduction
        • Tutorials
        • Small Molecule Modeling - Documentation
        • Small Molecule Modeling Release Notes
        • Small Molecule Modeling Release Notes
      • AI Fold Floes
        • Tutorials
        • How-To Guides
        • AI Fold Floes - Documentation
        • AI Fold Floes Release Notes
        • Release Notes
      • Format Conversion
        • orion-etl - Documentation
        • Format Conversion (ETL) Floes
        • Legacy Release Notes
      • Utility Floes
        • Introduction
        • Tutorials
        • How To Guides
        • Utility Floes - Documentation
        • Utility Floes Release Notes
        • Utility Floes Release Notes
        • Historical Release Notes
  • Antibody Discovery Suite
    • 3D Antibody Modeling Package
      • 3D Antibody Modeling Tutorial
        • Floes Used in This Tutorial
        • Importing Sequences
        • Antibody Model Generation
        • Modeling Results
      • Antibody MD Simulations Tutorial
        • Antibody Molecular Dynamics Simulations
      • Understanding Antibody Surfaces and Annotations
        • Surface Patches
        • Structure Annotations
      • Antibody SiteHopper-based Clustering Tutorial
        • Result Analysis
      • OpenEye 3D Antibody Modeling - Documentation
        • Antibody CDR Loop Ensemble Generation
        • Antibody Experimental Structure Prep
        • Antibody MD Simulations with Conformational Analysis
        • Antibody Mutate Residue(s)
        • Antibody SiteHopper-based Clustering
        • Antibody Structure Relaxation
        • Antibody Subset System
        • Antibody Surface Patch Generation
        • Antibody-Antigen Interface Ala-scanning
        • Export Dataset to Fasta
        • Humanize Fv Antibodies
        • Import Antibody FASTA Files
        • Renumber and Restyle Antibody
        • Sequences to 3D Models for Fv Antibodies
        • Sequences to 3D Models for Nanobodies
      • Release Notes
        • v0.1.8 July 2024
        • v0.1.3 April 2024
      • Bibliography
    • AbXtract™ - NGS Antibody Discovery
      • Introduction to AbXtract
      • Tutorials
        • Tutorial 1: NGS Pipeline with Custom Interactive Selection (PacBio), In-Vitro Library
        • Tutorial 2: NGS Pipeline with Automated Top Lead Selection (PacBio), In-Vitro Library
        • Tutorial 3: NGS and Sanger Pipeline with Automated Top Lead Selection (PacBio), In-Vitro Library
        • Tutorial 4: NGS and Sanger Pipeline with Automated Top Lead Selection (Illumina), In-Vitro Library
        • Tutorial 5: NGS Pipeline with Automated Top Lead Selection (PacBio), Patient Library
        • Tutorial 6: NGS UMIs Extract and Annotation Floe
        • Video Tutorials Featuring AbXtract
      • How to Guides
        • How to Use the Floe Report and NGS Select to Select Population of Interest
        • How to Use the Floe Report and SANGER Select to Pick Population of Interest
        • How to Condense a Dataset with Too Many Records
        • I Performed Enrichment. How Do I Know which Population of My Given Sequence was Found?
        • How to Reduce the Number of Fields in the Output
        • How to Download the CSV of the Dataset of Records
        • How to Visualize Custom Experimental Metrics
        • How to upload data from an AIRR-compatible file
        • How to export any dataset to an AIRR-compatible file
      • Frequently Asked Questions
        • Frequently Asked Questions about AbXtract Floes
      • Abxtract Floes - Documentation
        • AbXtract/AIRR File to Orion Dataset - AbXtract
        • Archive To Files
        • Automated Top Lead Selection - AbXtract
        • Cluster (AbScan) Antibody Binding Regions - AbXtract
        • Condense Dataset by Region of Interest by Most Abundant - AbXtract
        • Convert floe report to html files
        • Custom NGS Select by Seq ID of Additional NGS Representatives by Group - AbXtract
        • Custom SANGER Select of Additional NGS Representatives by Group - AbXtract
        • Export AIRR Fields for Dataset - AbXtract
        • Liability Quantification Across CDRs - AbXtract
        • Logomaker for Antibody CDRs - AbXtract
        • Modify Sample Name/Barcode Group for Downstream Processing - AbXtract
        • Multiple Round Relative Abundance and Enrichment Calculation by Region of Interest (ROI) - AbXtract
        • NGS IgMatcher, Annotation Only - AbXtract
        • NGS Pipeline - AbXtract
        • NGS Pipeline Efficiency - AbXtract
        • NGS Pipeline with Automated Top Lead Selection - AbXtract
        • NGS UMIs Extract and Annotation - AbXtract
        • NGS and Sanger Pipeline - AbXtract
        • NGS and Sanger Pipeline with Automated Top Lead Selection - AbXtract
        • Overlap Among Different Datasets - AbXtract
        • Quick Sanger from ABI Traces - AbXtract
        • Quick Sanger from DNA or Amino Acid Sequence Files - AbXtract
        • Single Round Relative Abundance and Enrichment Calculation by Region of Interest (ROI) - AbXtract
        • Subset the Number of Fields for Export - AbXtract
      • Key Fields
        • Liability Metric Fields
        • Biophysical Metric Fields
        • Identifier Fields
        • Overlap Fields of NGS to SANGER or NGS
        • Enrichment, Abundance and Relative Abundance Fields
        • Scaffold / Germline Call Fields
        • Clustering Fields
        • Annotation Fields
        • Sequence Quality Fields
        • Special Fields to Add to Upload (Use in Analyze Tool Only)
        • AIRR Fields
      • Release Notes
        • v0.1.9 July 2024
        • v0.1.6 April 2024
        • v0.1.5 October 2023
        • v0.1.4 December 2022
        • v0.1.3 June 2022
        • v0.1.2 April 2022
        • v0.1.1 March 2022
        • v0.1.0 December 2021
    • AI Fold Floes
      • Tutorials
        • Sequence-to-Structure Tutorial
        • Sequence-to-Structure with Ligand Affinity Ranking Tutorial
      • How-To Guides
        • How-To Guides for MSA Search Floes
        • How-To Guides for Structure Prediction Floes
      • AI Fold Floes - Documentation
        • MSA Align and Search
        • MSA Collection Setup from FASTA
        • Protein Sequence to AI Folded Structure Ligand Affinities
        • Protein Sequence to AI Folded Structure Prediction
      • AI Fold Floes Release Notes
        • v0.2.6 April 2026
      • Release Notes
        • v0.2.0 December 2025
        • v0.1.1 April 2024
        • v0.1.0 February 2024
    • Format Conversion
      • orion-etl - Documentation
        • Archive Import
        • Dataset Copy
        • Dataset to Collection Export
        • Dataset to File Export
        • Dataset to Record File Export
        • File to Dataset Import
        • File(s) and/or dataset(s) to archive
        • Record Collection to Dataset Import
        • Record File to Dataset Import
        • URL to File Import
      • Format Conversion (ETL) Floes
        • v6.7.0 September 2025
      • Legacy Release Notes
        • v6.6.0 August 2025
        • v6.3.2 January 2025
        • v6.3.1 September 2024
        • v6.3.0 August 2024
        • v6.2.0 June 2024
        • v6.1.2 February 2024
        • v6.0.0 September 2023
        • v2.3.0 July 2023
        • v2.1.3 February 2023
        • v2.1.2 November 2022
        • v2.1.1 September 2022
        • v2.1.0 July 2022
        • v2.0.2 February 2022
        • v2.0.1 December 2021
        • v2.0.0 November 2021
        • v1.2.9 November 2021
        • v1.2.8 October 2021
        • v1.2.7 October 2021
        • v1.2.6 June 2021
        • v1.2.5 June 2021
        • v1.2.4 November 2020
        • v1.2.3 August 2020
        • v1.2.2 April 2020
        • v1.2.1 March 2020
        • v1.2.0 February 2020
        • v1.1.1 October 2019
        • v1.1.0 August 2019
        • v1.0.0 July 2019
        • v0.1.29 April 2019
        • v0.1.28 February 2019
        • v0.1.27 February 2019
        • v0.1.26 February 2019
        • v0.1.25 January 2019
        • v0.1.24 December 2018
        • v0.1.23 November 2018
        • v0.1.22 November 2018
        • v0.1.21 October 2018
        • v0.1.20 October 2018
        • v0.1.19 October 2018
        • v0.1.18 September 2018
        • v0.1.17 September 2018
        • v0.1.16 September 2018
        • v0.1.15 September 2018
        • v0.1.14 September 2018
    • Utility Floes
      • Introduction
      • Tutorials
        • Spruce Prep Tutorial
        • Creating and Applying Molecule Filters
      • How To Guides
        • How to Guides for Spruce Floes
      • Utility Floes - Documentation
        • Build Sidechains
        • Calculate Dipole Moment
        • Cap Chain Breaks
        • DU to Mol
        • DU to mmCIF/PDB File
        • Dataset Append – Generating SMILES Field
        • Dataset Deduplication – Based on Molecule, String, Integer, or Float Field
        • Dataset Deduplication – Merge
        • Dataset Filtering – Create Custom Filter
        • Dataset Filtering – Custom or Built-in Filter Types
        • Dataset Manipulation – Add Molecule Title Field
        • Dataset Manipulation – Add Title to Molecule Field
        • Dataset Manipulation – Concatenation
        • Dataset Manipulation – Field Rename
        • Dataset Manipulation – Field Type Conversion
        • Dataset Subsetting – Random Splitting Or Selection
        • Dataset Subsetting Based on Dataset, Numerical, String, or Regex Field
        • Dataset Subsetting Based on String Keys
        • Extract Biological Units
        • Generate 2D Similarity Matrix
        • Generate 3D Similarity Matrix
        • Generate Fingerprints
        • Generate User-Defined Fingerprints
        • Generate and Deduplicate SMILES for One or More Datasets
        • Minimize Design Unit
        • Mutate Residue(s)
        • OMEGA - 3D Conformer Ensemble Generation
        • OMEGA - Generate a Single 3D Conformer
        • Protein Loop Modeling or Re-modeling
        • Protonate DU and structures
        • QUACPAC - Partial Charges
        • ROCS, FastROCS - Import Shape Query to Record
        • Receptor In DU
        • Residue State Changer
        • Rotamers of a Residue
        • SPRUCE - Import Prepared PDB Files
        • SPRUCE - Protein Preparation
        • Sample Collection By Shards
        • Subset Design Unit
        • Subset Design Unit Within
        • Subset Design Unit to Smallest Binding Unit
        • Substructure Search - Small Scale Substructure Matching
        • Superpose DUs
        • Swap Metal(s)
        • Update DU Content
      • Utility Floes Release Notes
        • v2.1.7 April 2026
      • Utility Floes Release Notes
        • v2.1.4 December 2025
        • v2.1.2 September 2025
        • v2.1.1 August 2025
        • v2.0.1 February 2025
      • Historical Release Notes
        • Biomolecular Modeling Floes Historical Release Notes
        • Cheminformatics Floes Historical Release Notes
        • Classic Floes Historical Release Notes
  • Partner Modules
    • The Gaussian Module
      • Gaussian Documentation
        • Gaussian Floe Tutorials
        • Frequently Asked Questions
        • OpenEye QM Gaussian Floes - Documentation
        • Gaussian Orion Module Release Notes
        • Legacy Release Notes
      • Format Conversion
        • orion-etl - Documentation
        • Format Conversion (ETL) Floes
        • Legacy Release Notes
    • CCDC GOLD Module
  • Cube Libraries
    • Molecular Dynamics Core Package
      • Installation
      • orionmdcore - Cube Documentation
        • Flask Preparation
        • Force Field
        • IO
        • MD Simulations
        • Simulation Flask Preparation
      • MD DataRecord
        • MD DataRecord a brief overview
        • MDDataRecord API Documentation
      • Molecular Dynamics Core Package
        • v2.6.1 April 2026
      • Legacy Release Notes
        • v2.5.5 September 2025
        • v2.5.4 July 2025
        • v2.5.3.1 September 2025
        • v2.5.3 March 2025
        • v2.5.2 March 2025
        • v2.5.1 February 2025
        • v2.4.1 June 2024
        • v2.4.0 April 2024
        • v2.3.1 February 2024
        • v2.1.2 September 2023
        • v2.1.1 September 2023
        • v2.1.0 July 2023
        • v2.0.1 April 2023
        • v1.1.6 June 2022
        • v1.1.5 April 2022
        • v1.1.2 December 2021
    • Snowball - The Primary OpenEye Cube Library
      • Definitions of Floe Classifications
      • snowball - Cube Documentation
        • 2D Depiction
        • 2D Similarity
        • 3D Similarity
        • Cheminformatics
        • Conformations
        • Debugging
        • Examples
        • Floe Report
        • IO
        • Lead Optimization
        • Library Design
        • Ligand Modeling
        • Ligand Preparation
        • Logic
        • MMDS
        • Models
        • Protein Ligand
        • SiteHopper
        • Spruce
        • Structure Based Design
        • Surface
        • Unclassified
        • Utility
      • Snowball Release Notes
        • v0.31.4 April 2026
      • Legacy Release Notes
        • v0.31.1 December 2025
        • v0.30.2 Sept 2025
        • v0.30.1 August 2025
        • v0.29.1 February 2025
        • v0.28.0 July 2024
        • v0.27.0 February 2024
        • v0.26.1 September 2023
        • v0.26.0 July 2023
        • v0.25.3 December 2022
        • v0.24.1 September 2022
        • v0.24.0 July 2022
        • v0.23.1 December 2021
        • v0.20.1 June 2021
        • v0.20.1 November 2020
        • v0.19.3 August 2020
        • v0.19.1 August 2020
        • v0.18.2 April 2020
        • v0.17.2 November 2019
        • v0.16.6 September 2019
        • v0.16.0 July 2019
        • v0.15.0 June 2019
        • v0.14.0 June 2019
        • v0.13.6 March 2019
        • v0.13.4 November 2018
        • v0.13.3 September 2018
        • v0.13.0 August 2018
  • OpenEye Glossary of Terms
  • Legal Notices
    • Copyright and Trademarks
    • Sample Code
    • Citation
      • Orion®
      • Orion Floes
      • Toolkits and Applications
      • Publications for Bibliographies
        • Orion
        • AFITT and FLYNN Applications
        • OEDocking Application and Toolkit
        • OMEGA Application and Toolkit
        • ROCS Application
      • OpenEye MMDS Web Service
    • Technology Licensing
    • GCC
      • GCC RUNTIME LIBRARY EXCEPTION
      • GNU GENERAL PUBLIC LICENSE
Orion Workflows
  • All OpenEye Documentation »
  • Contents »
  • Small Molecule Discovery Suite »
  • Hit Identification Module »
  • ROCS X Floes »
  • ROCS X Floes - Documentation »
  • ROCS X - Initialize 3D Search

ROCS X - Initialize 3D Search

Description

This floe initializes a ROCS X model for running a 3D similarity search on a query. First, the floe reads in a ROCS X 3D library that contains 3D conformers for synthons and a sample of products. An initial FastROCS search is performed with the product sample against the query. The hit list of this initial FastROCS search is used to seed the model. The top-scoring products in the hit list are fragmented to construct “bandit arms.” These bandit arms form the decision space that’s used when running a 3D search with the model. A second FastROCS search is performed with the bandit arms as queries against the synthon conformers to associate synthons with the bandit arms.

Key Inputs and Outputs

The key inputs are the query to search (only single-conformer molecule queries are supported at the time) and the ROCS X 3D library to search (typically the output from the Multi-Stage ROCS X Preparation or ROCS X - Prepare 3D Library Floe). A hit list from an external search on the query can be used as an optional input (see the Speed Up Preparation of a Very Large ROCS X 3D Library How-To Guide in the documentation).

The key output is a ROCS X 3D Search Model, which is typically used in the ROCS X - Run 3D Search Floe to run a 3D similarity search on the query with the model.

Cost Considerations

The floe cost scales with two key parameters: the number of synthons in the input ROCS X 3D Library and the Maximum Number of Bandit Arms parameter. The computational bottleneck is typically the FastROCS search of the bandit arm queries against the synthon conformers in the Parallel FastROCS (Synthons) Cube, which runs on a GPU instance. If the Maximum Number of Bandit Arms is half as much (500 compared to the default 1,000, e.g.), there will be half as many queries in the FastROCS search, so the cube should take half as long to run. Using too few bandit arms in the model can degrade search results. It is recommended to keep the Maximum Number of Bandit Arms at the default unless you are running a quick search or debugging.

Promoted Parameters

Title in user interface (promoted name)

Inputs

Query to Search (query_in): The dataset containing the query. Only single-conformer molecule queries are supported at the time. If a multi-conformer molecule is provided, the floe will immediately fail.

  • Required

  • Type: data_source

ROCS X 3D Library to Search (fastrocs_collection_in): The collection containing the 3D library to search.

  • Required

  • Type: collection_source

Initialize with Sampled Products (sample_products_flag):

  • Type: boolean

  • Default: True

  • Choices: [True, False]

Inputs: Model Initialization

External Hit List for Query (records_in): (Optional) Dataset containing a hit list of ROCS scores from an external search on the query.

  • Type: data_source

External Hit List Score Field (ext_hitlist_field_name): (Optional) Field for 3D similarity scores on the external hit list.

  • Type: field_parameter::any

  • Default: Tanimoto Combo

Outputs

ROCS X 3D Search Model Collection (collection_out): The name of the output collection containing the initialized model that can be run in a 3D search.

  • Required

  • Type: collection_sink

  • Default: ROCS X 3D Search Model

Hit List from Product Sample Dataset (initial_sample_hit_list_dataset): The name of the output hit list dataset from ROCS rescoring of the top hits in the initial FastROCS search of the product sample against the query.

  • Required

  • Type: dataset_out

  • Default: Product Sample Hit List

Temporary Collection (temporary_collection): This collection is created by the floe for internal use during the run and is automatically deleted by the floe when it finishes.

  • Required

  • Type: collection_sink

  • Default: Temporary Collection

Failures Collection (failures_out): The name of the output collection of failures.

  • Type: collection_sink

  • Default: ROCS X Initialize 3D Search Failures

Options: Basic Bandit Arm Search

Maximum Number of Bandit Arms (max_init_arms): Bandit arm creation will stop if the number of bandit arms exceeds this threshold. The cost of the floe scales with this parameter. If it is half as much (500 compared to the default 1,000, for example), the cube that is the computational bottleneck should take half as long to run. Using too few bandit arms in the model can degrade search results. It is recommended to keep this at the default unless you are running a quick search or debugging.

  • Type: integer

  • Default: 1000

Number of Hits to Fragment (num_fragment): The number of top-scoring hits that will be used to create bandit arms for the model.

  • Type: integer

  • Default: 2500

Clustering Score Threshold (cluster_threshold): Threshold ROCS score for bandit arm clustering. Fragments scoring higher than the threshold will be clustered together in a bandit arm. Setting this higher may result in more bandit arms being discovered.

  • Required

  • Type: decimal

  • Default: 1.6

Synthon Search Score Threshold (searchlist_threshold): Threshold ROCS score for synthon search. Synthons scoring higher than the threshold will be associated with the search lists for the bandit arms. It is recommended to set this slightly lower than the Clustering Score Threshold.

  • Required

  • Type: decimal

  • Default: 1.4

Synthon Search List Size (num_searchlist): Maximum number of synthons to store on bandit arm search lists.

  • Type: integer

  • Default: 10000

Options: Overlay

FastROCS Similarity Type (fastrocs_similarity_type): Type of FastROCS similarity to use to rank molecules for the FastROCS hit lists.

  • Type: string

  • Default: Tanimoto Combo

  • Choices: [‘Tanimoto Combo’, ‘Ref Tversky’, ‘Fit Tversky’, ‘Shape Tanimoto’, ‘Shape Ref Tversky’, ‘Shape Fit Tversky’]

ROCS Score Field (score_field): Field on the record that contains ROCS scores against the query.

  • Required

  • Type: field_parameter::float

  • Default: Tanimoto Combo

  • Choices: [‘Tanimoto Combo’, ‘Tversky Combo’]

Number of FastROCS Random Starts (number_of_fastrocs_random_starts): If specified, FastROCS will use the specified number of random starting orientations for each conformer being overlaid with FastROCS. If unspecified, the default of 4 inertial starts will be used. Compute time and cost scale roughly linearly with the number of starts.

  • Type: integer

Shape Only FastROCS Overlay (shape_only_fastrocs_overlay): If set to On, FastROCS will overlay molecules using shape only, ignoring color. If set to Off, FastROCS will overlay molecules using shape and color. Note that this parameter affects the overlay process but not the scoring; that is, the overlay can be done with shape while the scoring is done with shape and color.

  • Type: boolean

  • Default: False

  • Choices: [True, False]

Options: Re-scoring

Hit List Size (hit_list_size): Size of product sample hit list. The maximum allowed value is 100,000. The minimum allowed value is 1000.

  • Required

  • Type: integer

  • Default: 10000

Number of Molecules to Rescore (number_of_molecules_to_rescore): The number of top-scoring molecules from FastROCS that will be sent to any of the enabled post-processing methods (ROCS). Note that the outputted hit lists will still be of the size specified by the Hit List Size parameter, which is generally smaller than this number. Maximum allowed value is 100,000,000. Minimum allowed value is 100,000.

  • Type: integer

  • Default: 100000

Maximum Conformers (rescore_max_confs): Maximum number of conformations to generate for ROCS rescoring.

  • Required

  • Type: integer

  • Default: 50

Number of ROCS Rescoring Random Starts (number_of_rocs_rescoring_random_starts): If specified, ROCS rescoring will use the specified number of random starting orientations for each conformer being overlaid. If unspecified, the default of 4 inertial starts will be used. Compute time scales roughly linearly with the number of starts.

  • Type: integer

GPU Hardware

FastROCS Instance Type (fastrocs_instance_type): The instances excluded by default are known to be not cost-effective for FastROCS.

  • Type: string

  • Default: !g4dn.metal,!g5.12xlarge,!g5.24xlarge,!g5.48xlarge,!g4dn.12xlarge,!g3s.,!p3.

Spot Instance Policy for FastROCS GPU Instance (spot_instance_policy_for_fastrocs_gpu_instance): To run on SPOT instances, use the default setting of ‘preferred’. To run on ON-DEMAND instances, set the value to ‘prohibited’. ON-DEMAND instances typically cost 3–4 times more than SPOT instances, but are more available than SPOT instances when overall demand for GPUs on AWS is high.

  • Type: string

  • Default: Preferred

  • Choices: [‘Allowed’, ‘Preferred’, ‘NotPreferred’, ‘Prohibited’, ‘Required’]

Parallel Limit for FastROCS cubes (maxparallel): The maximum number of concurrently running copies of this cube.

  • Type: integer

  • Default: 100

Options: Advanced Bandit Arm Search

Color Force Field (color_force_field): Color force field to be used

  • Type: string

  • Default: ImplicitMillsDean

  • Choices: [‘ImplicitMillsDean’, ‘ExplicitMillsDean’, ‘ImplicitMillsDeanNoRings’, ‘ExplicitMillsDeanNoRings’]

Custom Color Force Field File (custom_cff_in): Custom color force field to use for FastROCS/ROCS overlays in the 3D search.

  • Type: file_in

Primary Connector Color FF Weight (weight1): The weight parameter (usually negative) for the color force field of the connector atom(s).

  • Type: decimal

  • Default: -1.0

Primary Connector Color FF Radius (radius1): The radius parameter for the color force field of the connector atom(s).

  • Type: decimal

  • Default: 10.0

Secondary Connector Color FF Weight (weight2): The weight parameter (usually negative) for the color force field of the connector atom(s).

  • Type: decimal

  • Default: -1.0

Secondary Connector Color FF Radius (radius2): The radius parameter for the color force field of the connector atom(s).

  • Type: decimal

  • Default: 10.0

Connector Distance Threshold (connector_dist_thr): Threshold distance (in Angstroms) for bandit arm clustering. If the distance between connector atoms is greater than the threshold, the fragments will not be clustered together in a bandit arm.

  • Required

  • Type: decimal

  • Default: 3.0

ROCS Start Type (start_type): The type of starting orientations for ROCS calculations.

  • Type: string

  • Default: Inertial

  • Choices: [‘Inertial’, ‘Rocs’, ‘AsIs’, ‘AtAtom’, ‘Random’]

Number of Random Starts (num_rand_starts): If specified, FastROCS will use this number of random starts for each conformer overlay. If unspecified, the default inertial starts will be used.

  • Type: integer

FastROCS Records Per Shard (fastrocs_records_per_shard): The target number of records in a FastROCS shard. The recommended default is 800,000.

  • Type: integer

  • Default: 800000

Output Fields

Overlay Molecule Field (overlay_molecule_field): Field on the output records that will hold the structure of the molecule overlaid by ROCS or FastROCS.

  • Type: field_parameter::mol

  • Default: Overlay Molecule

Tanimoto Combo Field (tanimoto_combo_field): Output field with the Tanimoto Combo. This field will only be created if the score type for FastROCS Similarity Type is Tanimoto Combo. The value in this field is a duplicate of the value in Combo Similarity.

  • Type: field_parameter::float

  • Default: Tanimoto Combo

Tanimoto Color Field (tanimoto_color_field): Output field with the Color Tanimoto. This field will only be created if the score type for FastROCS Similarity Type is Tanimoto Combo. The value in this field is a duplicate of the value in Color Similarity.

  • Type: field_parameter::float

  • Default: Color Tanimoto

Tanimoto Shape Field (tanimoto_shape_field): Output field with the Shape Tanimoto. This field will only be created if the score type for FastROCS Similarity Type is Tanimoto Combo. The value in this field is a duplicate of the value in Shape Similarity.

  • Type: field_parameter::float

  • Default: Shape Tanimoto

Tversky Combo Field (tversky_combo_field): Output field with the Tversky Combo. This field will only be created if the score type for FastROCS Similarity Type is Fit Tversky or Ref Tversky. The value in this field is a duplicate of the value in Combo Similarity.

  • Type: field_parameter::float

  • Default: Tversky Combo

Tversky Color Field (tversky_color_field): Output field with the Color Tversky. This field will only be created if the score type for FastROCS Similarity Type is Fit Tversky or Ref Tversky. The value in this field is a duplicate of the value in Color Similarity.

  • Type: field_parameter::float

  • Default: Color Tversky

Tversky Shape Field (tversky_shape_field): Output field with the Shape Tversky. This field will only be created if the score type for FastROCS Similarity Type is Fit Tversky or Ref Tversky. The value in this field is a duplicate of the value in Shape Similarity.

  • Type: field_parameter::float

  • Default: Shape Tversky

Best Query Field (best_query_field): Output field for the query with the highest similarity to the fit molecule.

  • Type: field_parameter::mol

  • Default: Query

Best Query ID Field (best_query_id_field): Output field for the ID of the query with the highest similarity to the molecule. This identifier will also appear in the dataset specified by the Output Query Dataset parameter.

  • Type: field_parameter::int

  • Default: Query ID

Previous Next

© Copyright 2026, Cadence Design Systems, Inc. Last updated on Apr 03, 2026.